Hepatitis B Virus Capsid Assembly Is Enhanced by Naturally Occurring Mutation F97L

In chronic hepatitis B virus (HBV) infections, one of the mostcommon mutations to the virus occurs at amino acid 97 of thecore protein, where leucine replaces either phenylalanine orisoleucine, depending on strain. This mutation correlates withchanges in viral nucleic acid metabolism and/or secretion. Wehypothesize that this phenotype is due in part to altered coreassembly, a process required for DNA synthesis. We examinedin vitro assembly of empty HBV capsids from wild-type and F97Lcore protein assembly domains. The mutation enhanced both therate and extent of assembly relative to those for the wild-typeprotein. The difference between the two proteins was most obviousin the temperature dependence of assembly, which was dramaticallystronger for the mutant protein, indicating a much more positiveenthalpy. Since the structures of the mutant and wild-type capsidsare essentially the same and the mutation is not involved inthe contact between dimers, we suggest that the F97L mutationaffects the dynamic behavior of dimer and capsid.

INTRODUCTION

Top

Introduction

The capsid of hepatitis B virus (HBV), the protein shell ofthe virus core, plays vital roles in nucleic acid replicationand intracellular transport (2, 3, 12, 14, 17, 18, 21, 27, 29).In vivo, the predominant form of the capsid is an $~$ 30-nm particle(24), corresponding to a T=4 icosahedron constructed from 240copies of the core protein (Cp), arranged as 120 dimers (9).The full-length Cp is 183 residues long, of which the N-terminal149 residues are the assembly domain (4, 33, 39), referred toas Cp149-wt in this paper. Cp dimers are the fundamental subunitof assembly (16, 22, 33, 39). In Cp149-wt, dimers are held togetherby a four-helix bundle with two helices from each half dimer(6, 8, 34). This four-helix bundle forms the spikes that decoratethe surface of HBV (Fig. 1).

View larger version (93K):
[in this window]
[in a new window]
FIG. 1. Residue 97 is largely buried in the middle of the HBV Cp dimer interface. (A) Surface representation of a T=4 HBV capsid shows 120 spikes (reproduced with permission from the publisher of reference 43). The spikes are formed by the four-helix bundle (B) at the dimer interface. Residue 97, located in the middle of the spike, is remote from the sites of interaction with other subunits (around the C-terminal peptide, to the right and left of the illustration) and nucleic acid (interior base of the four-helix bundle). Coordinates are from the crystal structure of a T=4 capsid of the ayw strain of HBV (34); residue I97 is shown as a space-filling model. (C) The amino acids surrounding I97 (boxed area in panel B) are shown in greater detail. I97 (green) is surrounded by hydrophobic residues (V27, L31, L55, I59, C61, W62, M93, G94, L100, and L101); an exterior edge of the pocket is bordered by a salt bridge between E64 and K94. The ß methyl of isoleucine (also in strains adr and adw) or the bulk of phenylalanine (in strains adyw and ady) at position 97 probably constrains movement of the side chain within the pocket (or, conversely, constrains movement of the pocket). However, leucine at position 97 would be expected to have greater freedom of movement. Backbone atoms are not shown, carbon atoms are gray, nitrogen atoms are blue, oxygen atoms are red, and sulfur atoms are yellow. The ribbon diagram was prepared with WebLab (Molecular Simulations, Inc., San Diego, Calif.).

Mutation of amino acid 97 to leucine, one of the most commonmutations of the core protein seen in chronic infections (13),has been found to affect nucleic acid metabolism (26, 35-37).Depending on the strain of HBV, residue 97 is phenylalanine(e.g., ayw and adyw) or isoleucine (e.g., adr and adw). In tissueculture studies with both F97L and I97L mutants, a substantialfraction of secreted capsid particles contained single-strandedDNA, instead of the mature form of the genome (36, 37). In theHuh7, but not the HepG2, cell line I97L correlates with increasedreplication compared to that of the wild-type virus (26).

Residue 97 is located in a hydrophobic pocket in the middleof the four-helix bundle with most of its surface buried (Fig.1). Based on the structure of the capsid (34), it is not clearwhy a conservative mutation in the dimer interface should changethe behavior of the virus. The site of the Cp149-F97L mutationis remote from important sites of interactions between dimers.The site for intersubunit contact is dominated by residues 125to 142, at either end of the dimer, $~$ 30 Å distant (Fig.1). The base of the dimer interface, about 20 Å away,forms the interior of the virus and may contact encapsidatednucleic acid and reverse transcriptase. The outer extremityof the four-helix bundle is believed to participate in bindingsurface protein to mediate export (5, 10, 28), though it hasalso been observed that mutations at residues 95, 96, and 97affect export (15, 20, 36).

In our previous studies, we described the assembly thermodynamicsand kinetics of Cp149-wt from the adyw strain of HBV. Assemblywas fast and nucleated and showed few intermediates (7, 44).The individual subunit-subunit interactions were very weak,on the order of –3 to –4 kcal/mol (depending onreaction conditions), which corresponded to a millimolar associationconstant per contact (7). The resulting capsids were stablebecause each subunit was involved in four contacts and becauseof a kinetic barrier to dissociation (hysteresis) (22). Theionic strength dependence of assembly and the effect of Zn²⁺on assembly kinetics suggested that assembly was regulated bya transition of Cp from the assembly-inactive state to the assembly-activestate (7, 25).

In this study we observe that the Cp149-F97L mutant proteinassembles faster and with dramatically different thermodynamicsthan Cp149-wt, suggesting that this seemingly innocuous mutationenhances the ability of the capsid protein to adopt the assembly-activestate and pass through the transition state of the assemblyreaction.

MATERIALS AND METHODS

Top

Materials and Methods

Protein. Residue 97 was mutated from phenylalanine to leucine with theStratagene QuikChange kit. The plasmid construct pCp149 (33)was mutated with the mutagenic primer GTGGGCCTAAAGCTCAGACAATTATTGTGG,where the leucine codon is underlined. Cp149-wt and Cp149-F97Lwere expressed in Escherichia coli and purified as previouslydescribed (42), except that cells expressing F97L were grownat 30°C. The mutation was confirmed by DNA sequencing andmass spectrometry. An extinction coefficient ( ${varepsilon}$ ₂₈₀) of 60,900M^–1 cm^–1 (43) was used for both proteins.

Assembly and stability assays. The Cp dimer was dialyzed into assembly buffer (50 mM HEPES[pH 7.5], 1 mM dithiothreitol) prior to reactions. The dimerwas assembled by mixing it with an equal volume of 0.3 M NaClassembly buffer, to yield a final NaCl concentration of 0.15M. Assembly reactions were monitored by light scattering ina SPEX Fluoromax-2 fluorometer (Instruments S.A. Inc., Addison,N.J.) (44). Concentrations of capsid and dimer were determinedby size exclusion chromatography (SEC) using a 27-ml SephacrylS-300 HR column or a 21-ml Superose 6 HR column connected toan AKTA FPLC instrument or a Shimadzu high-pressure liquid chromatographyinstrument. The capsid eluted at 8 ml and the dimer eluted at18 ml in the Superose 6 column (7). Circular dichroism (CD)spectra were recorded at 21°C with a Jasco J-715 spectropolarimeterand a Hellma 0.1-mm path-length quartz cell (22). Multianglelaser light scattering data were measured with a Dawn detector(Wyatt Technology, Santa Barbara, Calif.).

Data analysis. The SEC data were analyzed to determine thermodynamic parametersas previously described (7). Briefly, for the assembly reactionwhere 120 dimers form a capsid, the equilibrium constant, K_capsid,is given by

(1)
where [dimer] and [capsid]are the concentrations of dimer and capsid, respectively. K_capsidincludes contributions from all 240 quasiequivalent interdimercontacts. The average per-contact association constant, K_contact(equation 2), is in the more convenient units of inverse molarity;the coefficient in equation 2 describes the degeneracy of theassembly reaction (11, 41, 46).

(2)
whereK_contact is related to the free energy per contact of each dimerby

(3)
where R is the gas constant andT is the temperature in degrees Kelvin.

RESULTS

Top

Results

Comparison of capsid stability determined from assembly reactions. SEC is a simple and direct assay to determine concentrationsof capsid and dimer at equilibrium. Capsid assembly reactionsclosely approach equilibrium, allowing determination of associationenergy (i.e., stability) for a given set of conditions (7, 11).No detectable intermediates were observed in chromatographsof either wild-type or mutant protein, as expected for a nucleatedreaction (44). SEC data also show the capsids and dimers ofthe wild type and mutant elute at the same volumes (Fig. 2A,inset). Multiangle laser light scattering data were consistentwith assignment of the two peaks as the 4-MDa capsid and 34-kDadimer (data not shown). In agreement with other reports (19),Cp149-F97L and -wt capsids were indistinguishable by negative-stainelectron microscopy and on sucrose gradients (data not shown).

View larger version (16K):
[in this window]
[in a new window]
FIG. 2. Mutation F97L alters assembly properties of Cp149. (A) A van't Hoff plot comparison of assembly of Cp149-F97L (dashed line) and Cp149-wt (solid line). The equilibrium data were obtained by SEC (the inset shows F97L [dashed line] and the wild type [solid line]). The per-contact association constants for both proteins at 21°C were identical (intersection of lines; see Table 1); however, the temperature dependence of the assembly reaction for Cp149-F97L (squares) was much greater than that for the wild type (circles). Data were fit to straight lines, imposing temperature independence on ${Delta}$ H, with slopes proportional to enthalpies of +7.4 kcal/mol for Cp149-F97L (dashed line) and +1.5 kcal/mol for the wild type (solid line). Each point is the average of at least seven measurements. (B) The assembly kinetics at 21°C of Cp149-F97L (squares) and Cp149-wt (circles) were monitored by 90° angle light scattering (in arbitrary units) at 320 nm. Assembly was induced by addition of salt to a final concentration of 0.15 M. Final protein concentrations were 20 (solid symbols) and 25 µM (open symbols). Though Cp149-wt was slower than Cp149-F97L, both reactions reached the same final concentration of capsid. (C) The time course of capsid assembly at 37°C was observed at 1-h intervals by SEC. The percentage of protein in capsid form for the mutant Cp149-F97L protein (squares) increased very rapidly during the first few hours of assembly, whereas the reaction for Cp149-wt (circles) was much slower. The data were fit to a hyperbolic curve.

Assembly reactions at 21°C showed that Cp149-wt and Cp149-F97Lassemble to the same extent but with different kinetics (Fig.2A and B). The effect of the mutation was even more evidentwhen the assembly reactions were compared at 37°C; the yieldof capsid was greater and assembly was much faster for Cp149-F97Lthan for Cp149-wt (Fig. 2C). Above 37°C, unlike Cp149-wt(7), Cp149-F97L had a marked tendency to form heterogeneousaggregates with masses on the order of tens of megadaltons,based on multiangle laser light scattering (data not shown).

Assembly of Cp149-F97L was measured at a range of protein concentrationsto establish isotherms for determining K_capsid, and at differenttemperatures to determine enthalpy ( ${Delta}$ H) and entropy ( ${Delta}$ S). Valuesfor K_capsid, the association constant for capsid from 120 dimers(equation 1), were obtained from reactions that were equilibratedfor 24 h. Longer incubations did not affect the yield of capsid(Fig. 2C). K_capsid was partitioned into K_contact (equation 2),the association constant for individual dimer-dimer interactions.The K_contact values of both proteins at 21°C were identical,corresponding to a ${Delta}$ G_contact value of –3.0 kcal/mol (equation3) or a dissociation constant of only $~$ 6 mM (Table 1). Becauseeach dimer makes quasiequivalent contacts with four neighboringdimers, this association energy results in a micromolar apparentdissociation constant for the capsid, where K_D_apparent is definedas the concentration where [dimer] and [capsid] are equal (7,40, 41). However, at 37°C the association energy for Cp149-F97Lwas –3.6 kcal/mol, substantially stronger than that forCp149-wt (Table 1). In short, the temperature dependence ofassociation is far greater for Cp149-F97L than for Cp149-wt.

View this table:
[in this window]
[in a new window]
TABLE 1. Thermodynamic parameters describing assembly of Cp149-wt and -F971L^a

The K_contact data were compared in a van't Hoff plot (Fig. 2A).The slope in a van't Hoff plot is equal to – ${Delta}$ H/R, whereR is the universal gas constant. The negative slopes in Fig.2A show that both reactions had a positive enthalpy of association.The greater temperature dependence for the assembly of Cp149-F97Lcompared to that for the assembly of Cp149-wt indicated a muchlarger enthalpy of assembly. Enthalpy was calculated as a temperature-independentparameter since our data were not sufficient to describe heatcapacity-induced curvature of the van't Hoff plot.

Assembly of Cp149-F97L was much faster than that of Cp149-wt. For a simple reaction, the initial rate of reaction is independentof the end point; i.e., kinetics are independent of associationenergy. However, capsid assembly is best described as a cascadeof reactions, where the stability of intermediates has a markedeffect on the rate of appearance of the final product (11, 41).Comparison of kinetics, showing that Cp149-F97L had a fasterintrinsic rate of assembly, was unambiguous at 21°C (Fig.2B), where both proteins have the same association energy.

Light scattering is a convenient method for measuring HBV capsidformation. SEC data demonstrate that concentrations of intermediatesin these reactions are very low so that the increase in lightscattering is almost exclusively due to accumulation of capsids(45). At 21°C and 150 mM NaCl, a concentration greater than15 µM Cp149-wt is required for appreciable accumulationof capsid (7). Under these conditions, Cp149-F97L assembledmuch faster than Cp149-wt, though the final yields of capsidwere the same (Fig. 2B). At 37°C, the greater stabilityof intersubunit contacts combined with faster kinetics furtherenhanced assembly of Cp149-F97L relative to that of Cp149-wt(Fig. 2C).

Comparison of the structure and stability of Cp149-F97L and -wt dimers. There are many possible explanations for enhanced enthalpy ofassembly for the mutant protein. One possibility is that themutant protein refolds during assembly. To test this possibility,we examined the CD spectra of both proteins and the sensitivityof the proteins to denaturation. The two proteins had nearlyidentical spectra, indicating that they have the same largely ${alpha}$ -helical secondary structure (Fig. 3A). We examined the stabilityof wild-type and mutant dimer by observing their denaturationin guanidine hydrochloride (GuHCl). We found that the GuHCltitrations of Cp149-wt and -F97L were nearly identical, indicatingthat the two proteins had approximately the same stability andthat both proteins were folded in the absence of denaturant(Fig. 3B). These results are in agreement with fluorescencestudies (data not shown).

View larger version (15K):
[in this window]
[in a new window]
FIG. 3. CD spectroscopy demonstrates that Cp149-F97L and -wt have similar structures and stabilities. (A) CD spectra recorded for Cp149-wt (solid line) and Cp149-F97L (dashed line). The double minima at 210 and 222 nm are diagnostic for ${alpha}$ -helical secondary structure. (B) Molar ellipticities at 222 nm, averaged from three or more CD spectra, of Cp149-wt (circles) and Cp149-F97L (squares) were plotted as a function of GuHCl concentration. CD is sensitive to unfolding at $≥$ 3 M GuHCl but insensitive to capsid dissociation (22). A single line fit both data sets, indicating that the wild-type protein and Cp149-F97L do not have significant differences in stability.

DISCUSSION

Top

Discussion

To better understand regulation of the life cycle of HBV, wecompared assembly of Cp149-wt with that of Cp149-F97L, whichhas a mutation that is common in chronic infections. We foundthat the Cp149-F97L mutant capsid protein assembles more rapidlythan Cp149-wt at 21°C, but to the same equilibrium concentration.At 37°C, Cp149-F97L assembly is faster and forms a thermodynamicallymore stable particle. The kinetic result implies that the mutantprotein is more able to assume the transition state conformationfor the assembly reaction, i.e., the activation energy ( ${Delta}$ G) islower for Cp149-F97L (Fig. 4). The enhanced stability of theCp149-F97L capsids may be due to changes in the stability ofthe unassembled form of core protein dimer and/or its assembledform in capsids. Effectively, the mutant protein is able tosupport capsid assembly at concentrations too low for the wild-typeprotein (Fig. 2C).

View larger version (23K):
[in this window]
[in a new window]
FIG. 4. Model describing the energetics of wild-type and Cp149-F97L capsid assembly. Our data show that at 21°C the K_contact values for both proteins are the same. However, the kinetics of capsid formation were much faster for the mutant than for the wild type. This indicates that the energetic barrier to reach the transition state must be smaller. Because the observed rate of assembly will also be affected by the concentration of the assembly-active state, the faster rate could be due to either a more stable active state of the dimer and/or a smaller ${Delta}$ G. The energy barrier represented by ${Delta}$ G in this figure includes nucleation and subsequent assembly. At physiological temperature, the ${Delta}$ G_contact of Cp149-F97L assembly is stronger and the rate constant is faster than those for Cp149-wt. Thus, at 37°C the Cp149-F97L dimer may start at a higher energy state than the wild type and/or the capsid may be at lower energy. The enhanced rate may be due to changes in the stability of the assembly-active state or ${Delta}$ G from that at 21°C. However, we observed that dimer stabilities for both proteins are about the same, and thus stability is not linked to the energy of assembly. Therefore, the difference in rate must be attributable to the ability of the mutant protein to undergo the conformational changes required for assembly. Thus, we propose that the Cp149-F97L mutation affects assembly by altering protein dynamics.

The increased stability of mutant capsids at 37°C was theresult of a dramatic change in ${Delta}$ H and ${Delta}$ S. This result is surprisingbecause the site of the F97L mutation is the intradimer four-helixbundle, not the interface between dimers. Generally, the enthalpyof a reaction correlates with buried surface area (1, 23), butthe site of the F97L mutation is not close to surfaces thatare buried during assembly (34). We eliminated the possibilitythat the enthalpy difference was due to partial unfolding ofCp149-F97L by showing stability similar to that of the wildtype based on denaturation studies (Fig. 3). Our experimentsalso showed that kinetics of Cp149-F97L assembly were faster(Fig. 2B and C), indicating a lower energy barrier to assembly(Fig. 4). These results are consistent with the hypothesis thatthe F97L mutation supports a conformational transition thatis involved in the assembly reaction.

Leucine has physical properties similar to those of isoleucineand phenylalanine but leucine is structurally distinct. Mutationof leucine to isoleucine can have drastic consequences for structure(see, for example, reference 32). In the HBV capsid structure(34) (Fig. 1) it can be seen that the ß methyl ofI97 constrains its freedom of motion. A phenylalanine at position97 would fill most of the pocket observed in the four-helixbundle, similarly constraining the residue and the protein.A leucine, which has no ß methyl and is smaller thatphenylalanine, would be expected to have significantly greaterfreedom of motion. Because there are (i) altered thermodynamics,(ii) enhanced kinetics, and (iii) presumably no change in theprotein surfaces that are actually buried during assembly, wepropose that the F97L mutation makes it easier for the dimerto undergo the structural transition required for assembly.

The large changes in enthalpy and entropy suggest significantstructural differences between Cp149-wt and -F97L, associatedwith the assembly reaction. The faster association rate indicatesthat this change also affects the stability of the assemblytransition state ( ${Delta}$ G in Fig. 4). However, we found no evidencefor structural differences between dimers (Fig. 3). The transitionstate may be related to the conformational change from an assembly-inactivedimer to an assembly-active state, hypothesized for capsid assembly(7, 25). The hypothesis that HBV assembly is sensitive to changesin protein conformation is consistent with our recent observationthat Zn²⁺ induces a conformational change in the unassembleddimer and increases the rate of capsid assembly without alteringassembly thermodynamics; i.e., Zn²⁺ changes the stability ofan intermediate state compared to reactant (dimer) but not thestability of the product (capsid) compared to reactant (25).We have suggested that formation of this intermediate-state,assembly-active dimer plays a role in regulating assembly (25).How does our study of the assembly of empty capsids comparewith assembly of cores in vivo? Regulation of capsid assemblyis probably a mechanism for limiting encapsidation of nonviralnucleic acid; otherwise one would expect a high proportion ofcapsids containing nonviral RNA. Thus, regulating factors andthe intrinsic association energy and kinetics are likely tobe critical to HBV assembly in vivo, though not necessarilyindependent of the Cp RNA-binding domain and viral RNA.

A recent report (19) observed that wild-type and I97L capsidsare equally stable, based on resistance to denaturants. We arguethat dissociation is an insensitive assay for determining capsidstability. Unlike capsid assembly reactions, which closely approachequilibrium (7, 11), dissociation reactions are affected byhysteresis (22, 30), i.e., there is a kinetic barrier to equilibrium.Based on thermodynamic-kinetic models of assembly (22), we haveshown that hysteresis occurs because the large intermediatesfound in the early stages of a dissociation reaction preferentiallyreassociate to form capsids, except when ${Delta}$ G_contact decreasescatastrophically, as if by a great excess of denaturant. Thislocal energy minimum impedes the reaction from equilibratingto the global energy minimum. Consequently, the stability ofviruses and other large oligomers can be determined only fromassembly reactions, not dissociation studies. Unlike largeroligomers, monomers and dimers generally do not display hysteresis(30), allowing us to compare stability of wild-type and F97Ldimers (Fig. 3B).

Mutation of core protein residue 97 is among the most commonmutations observed in chronic infections (13, 36). Why woulda mutation such as F97L be selected for? The enhanced associationenergy and kinetics of assembly indicate that F97L can supportassembly even when expression of the capsid protein is lower.The increase in intracellular replication (26) would be an expectedconsequence of increased capsid assembly, since capsid formationis believed to activate reverse transcription (2, 3, 12, 18,27, 29). Premature or immature secretion (36, 37) may also berelated to increased availability of core. Mutations to polymerase(31) or treatment with the polymerase inhibitor lamivudine (38)also leads to secretion of single-stranded DNA-containing particles,which suggests that there may not be a strict link between thecompletion of genome replication and core particle secretion(12). Thus, any mutation that increases the availability ofcores could increase the fraction of secreted immature particles,as DNA polymerization probably stalls when access to deoxynucleosidetriphosphates is prevented by envelopment.

ACKNOWLEDGMENTS

A research scholar grant (RSG-99-339-04-MBC) from the AmericanCancer Society supported this work.

We thank Jennifer Johnson and Christina Bourne for helpful discussionsand Bruce Baggenstoss at the EPSCOR Mass Spec facility for hisassistance.

FOOTNOTES

* Corresponding author. Mailing address: University of Oklahoma Health Sciences Center, Dept. of Biochemistry and Molecular Biology, P.O. Box 26901, BRC 464, Oklahoma City, OK 73190. Phone: (405) 271-9030. Fax: (405) 271-3910. E-mail: adam-zlotnick{at}ouhsc.edu.

REFERENCES

Top